INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

peria

-0.68

 Doctrine

-0.67

udder

-0.67

sup

-0.67

arenthood

-0.66

bell

-0.66

obyl

-0.66

forth

-0.65

erb

-0.64

etus

-0.62

POSITIVE LOGITS

 Hassan

0.67

 Isles

0.65

 executed

0.61

bom

0.60

isoft

0.59

wait

0.58

bee

0.58

issance

0.58

perm

0.57

 manif

0.56

Activations Density 0.000%

No Known Activations

This feature has no known activations.